Not the definition of composition-dependent
Splitting a pair of nines isn't my idea of composition-dependent BS. No one would consider 9-9 an 18. It belongs to the group of BS decisions we consider to be "the pair splits." To me, this isn't the same as c-d BS, where, when I have a 12, I worry about which cards I have in my hand. I hope you see the difference.
In any event, I'll go out on a limb and say the following: if you read it on Mike Shackleford's brilliant thewizardofodds.com site, it is correct. And, the fact that he agrees with Griffin cements the deal. Others who disagree? Too bad for them!
Don